Overhauling high voltage vacuum tube (valve) power supply

[proschuno]

So I"m overhauling a 50 year old home-builtPSU in serious need of refitting. Using two pass tubes of them has a busted filament and the other was replaced about a year ago. Should I worry about getting a matched pair or just get a single tube?

Also, the circuit uses two oil filled caps that I would like to replace for good measure (along with all the other ones in the schematic). Seeing that these on ebay are quite expensive along with any other electrolytic alternative, I'm wondering if I couldn't simply tie two 450V caps in series as shown for the +B supply? This would be much cheaper than getting another oil cap or similarly rated electrolytic ($US 80 on Mouser).. I notice one of the oil caps' leads is tied before the choke, so I wonder why so?
Proschuno

And for the missing piece: The OA2 tubes, along with the 1st and 3rd leads of the pot, are in parallel with the 30MF caps, and the cathode of the error amp tube is connected to the wiper of the pot.

And BTW, taking apart the PSU I found the schematic; notice the date…. it's literally as old as my mother lol  )

The pass tubes are 1619 (or VT164), the error amp is 6sj7, the rectifier for +B is 5v4g, and the 2nd one is 6x4.

[SeanB]

Just be careful of those oil filled caps, they likely contain PCB oil, and should only be handled with gloves, and if there is any oil leakage clean it off with solvent and dispose of the rags, gloves and caps as dangerous waste. You could replace them with 4u7 1kV film capacitors with no problems.

[calexanian]

The 8uf oil cap and that choke form a VERY important network!!!!!!!  That is a capacitor input to filter type circuit. Much of the ripple rejection of that supply comes from that combination of a low capacitance and the choke. Otherwise you will have a lot of ripple and the plate voltage will be way too high for the pass tubes to handle. Just bring the supply up slowly on a variac and I bet that oil cap will be fine. Monitor the current draw to the unit and with no tubes apart from the rectifier tube. Watch the voltage on that cap as well. As the filaments heat up in the rectifier it will self soft start so to speak.  Depending on age those electrolytics are toast. If the oil cal is bad, replace it with as close to 8uf as you can get.

[SeanB]

Just looking 4u7 630V caps are hard to get. You can use 2 parallel 2u2 630V film capacitor to replace them, they are $4 each. The electrolytics are a lot cheaper, you can use 33uf 450V Nichicons as replacements for the 47uF ones, they are under $2 each. If you want to use to replace the 4uF oil then use 10uF 450V capacitors with a 100k 1W resistor across each capacitor like the other side.

[proschuno]

Ok also too, my concern is wattage rating, if it should be one; is this also a reason oil caps were used too? Or can I literally replace them with the polypropylene/ethylene caps that I have, or are we talking some high power film caps?

[SeanB]

Oil filled was used because the old capacitors were paper insulation, which needed the oil fill to keep water and moisture out which would go bang. Any film capacitor rated at 630V DC or more will work as a replacement. Just put a 10R to 22R 1W resistor in series with the capacitor, as the originals would have had a high series resistance and the diode rectifier valve would have been designed around it.

[dannyf]

On high voltage caps: you can find them in camera flashes, or transformer-less smps (usually 470v or less, however).

I would explore if a smps can be made to produce the regulated voltage - should be much smaller, lighter and more efficient.

[N2IXK]

Agree with the comments on the oil capacitor. If it is bad/leaking, the replacement should be as close to the original value as possible. Excessive capacitance here will kill your rectifier tubes in short order.

As far as the pass tubes go, no need for specially matched pairs.  1619s haven't been manufactured in probably 50 years, so NOS ones from eBay or similar are your only option. Fortunately, even though they are 6L6 variants, the 2.5V directly heated cathode kept them from gaining cult status among the audio and guitar amp crowd. They are still pretty easy to get, and cheap compared to the rest of the 6L6 family.

[N2IXK]

On high voltage caps: you can find them in camera flashes...

Photoflash caps are intended for energy storage applications. They DO NOT like being used in ordinary filter applications, as they can not tolerate ripple current well, and have a very short life (often with rather spectacular failure mode) when used as such.

[proschuno]

I would explore if a smps can be made to produce the regulated voltage - should be much smaller, lighter and more efficient.

I actually have considered this for building one, but at high voltages I'm to inexperienced to build it, and I would rather have the low-noise and transient response of linear regulators. The device is already made, and trust me, at nearly 10lbs its a brick! And also let's try finding a switching transformer at those rated voltages….

But actually my main concern was the wattage rating for the oil caps… but they're not leaking so they seem fine.

[dannyf]

Maybe it is a long-term solution and you can start working on smaller boost-up converters first, with the goal to build your own sometime down the road.

Getting a transformer for a smps, or winding one, is not that difficult, especially if you don't need lots of power output.

Transient response could be an issue, however. If it is used to power tube stuff, they are likely class A so transient response isn't a big deal.

[calexanian]

The ripple in those input caps is HUGE in those circuits. It you are going film make sure you have a very low esr. Also I do not recomend series. Parallel is a great idea as it lowers esr
And distributes heat. Yes......   They do heat up in that kind of service! Those regulated circuits are kind of an animal all their own.. They however can be extremely reliable.

[T3sl4co1l]

FYI, the resonant frequency of that cap to the choke is 28Hz, so it's not a resonant filter.  It's just a really awful C-input filter.  This gives terrible supply resistance (VDC = 1.4 * Vrms for light loads, dropping to maybe 1.1x for... 30% loading maybe, down finally to ~0.85x for 60-100% loading), but the low capacitance is required to keep the rectifier happy while running near ratings.

Tube diodes cannot withstand large peak currents; typically the absolute maximum cap-input value is 40uF, and that's with derated current capacity.

The ripple through the capacitor is miniscule: we're talking 100s of mA here.  Okay, fairly large compared to the ratings of an electrolytic that size, but sheesh, you can put whole amperes through a film cap, no biggie.  The ripple *voltage* however is fairly large, much more than the 10-20% you'd want on an electrolytic.  The voltage swing will be the full peak voltage at heavy loads (where the diode current is continuous, and the filter is acting as a choke-input filter), but less at lighter loads (cap-input filter action).

In a regulated supply, I suppose the supply resistance doesn't matter much, so who cares.  But that said, I don't know why they didn't just leave it choke-input altogether, why bother with an input cap at all?  They certainly could've added a few more henries for the cost of that oil-can cap, probably getting better performance (less ripple, better regulation) in the end.

Tim

[calexanian]

Believe me. Put an ac meter in there and you will see much higher ripple than that! The resonance is supposed to be less than 120hz. Far less. It's not a dc filter in a traditional seance. It's to provide an ac component to the current flowing through the choke to provide proper filtering there. You just have to play with those circuits to understand. Look up the section in the RCA radiotron design handbook under power supplies. It's the best description and explanation of how those circuits work anywhere. You can see how they did it before our modern stuff. I highly recommend that book to everybody. Just the basic circuit theory in there is better than any modern textbook.

[oldway]

The ripple in those input caps is HUGE in those circuits. It you are going film make sure you have a very low esr. Also I do not recomend series. Parallel is a great idea as it lowers esr
And distributes heat. Yes......   They do heat up in that kind of service! Those regulated circuits are kind of an animal all their own.. They however can be extremely reliable.

Concern with ESR for an old valve/tube equipment looks very funny to me: in this time, whe did'nt even know what ESR was... 
In the sixties, i have build a lot of vacuum tube audio amplifiers and repair hundreds of vacuum tube tv's (black and white and colors tv's) and i had never hear somobody speaking about ESR.
Later, when I worked with high power fast SCR's inverter, concern was not to exceed RMS current in the electrolytics. It was low frequency ripple (120Hz/180Hz/360Hz)
ESR concern was important for SMPS because the high frequency ripple.

[T3sl4co1l]

Yes, ESR in those old caps (the few that survive without horrendous ESR!) was on the order of ohms.  Miliohms here or there on a film capacitor will go completely and utterly unnoticed.

As I expected, my intuition was in the right ballpark:



The cap's Irms = 158mA is quite reasonable.  108mA DC output at 432V, probably a bit over ratings (the choke says 100mA, but I don't know what it's actually expected to run at).

As it turns out, the capacitance is too large to fully transition to choke-input mode operation at rated current (that'll take more like 200-300mA).  So I'll admit to being wrong about that.  Accordingly, the output voltage is more like 1.0 x Vrms at rated current, up to 1.4x at no load (and following some curve inbetween).

134Vp-p is still kind of a lot of ripple voltage, more than an electrolytic would want to deal with.  So yeah, film is still a good idea.  One of those EPCOS "DC Link" caps would be more than enough

Note this simulation includes 200 ohms series resistance (representing the transformer and diodes' series resistance), and ideal capacitors and inductor.  200 ohms is probably being generous (that might be the transformer alone, not even including the 5R4/5V4), and the choke is probably 50 or 100 ohms again; actual output is probably more like 350V at the same load.

Tim

[proschuno]

BTW, the choke is 375 ohms (it says so on the label and it measures that.). So are these components purposely designed, or could I actually tweak this design to improve performance slightly I wonder? This was homebuilt, so I'm sure the guy who built it just threw together what he had on hand, so I could go back and make it better….

Second Question: Would these caps be suitable to replace the oil caps? I hooked one up to a 1khz square source on my oscilloscope in series with a resistor and it measured to be 4 microfarads, so I'll suppose the other is in the same condition (don't know if I stated this before, but noise, ripple, and nothing else like that was ever a problem with this. But in the event I would need to, I'm just curious, could the film caps below (attached photo) be used?

[T3sl4co1l]

Whatever it is, it has to be rated for 600VDC absolute minimum; I'd probably get 1000V caps myself.  Then, if the value is alright, doesn't matter much beyond that.

Tim

[oldway]

Why do you want to replace the oil caps ?
If it has no leakage, there is nothing wrong with this caps.
It is doing his job and will do it for 50 years more.

Never replace what is not really necessary.

If ripple is low, you don't have to worry about the caps.

You want to make better this power supply.

It is a very old design and has a lot of shortcomings.
To make it better, you should make a new one.
A new design would be a SMPS with power FET, voltage regulation and current limiting, digital readings, ...

For a sixties design, I think that is not a very good one:
- there would be several voltage ranges with transformer secundary taps comutations for reducing anode power dissipation.
- OA2 voltage stabilisators are not that good : noisy, not very stable, often oscillates, ...
- 5V4 (more serial resistance) and 6X4 could be replaced with  silicon rectifiers. In those times, I used BY100 silicon rectifiers.
- I would prefer to use EL34/6CA7 in place of 1619. If it would be a later design (seventies), i would choose to use high power color TV pentode PL/EL519.

An example of a far better design:


An other one with 7 ranges:

[dannyf]

For high impedance / low current devices like tubes, high ESR is either irrelevant or desirable.

[proschuno]

Believe me. Put an ac meter in there and you will see much higher ripple than that! The resonance is supposed to be less than 120hz. Far less. It's not a dc filter in a traditional seance. It's to provide an ac component to the current flowing through the choke to provide proper filtering there. You just have to play with those circuits to understand. Look up the section in the RCA radiotron design handbook under power supplies. It's the best description and explanation of how those circuits work anywhere. You can see how they did it before our modern stuff. I highly recommend that book to everybody. Just the basic circuit theory in there is better than any modern textbook.

For the radiotron designer's handbook, would this be what you're talking about?

http://www.tubebooks.org/Books/RDH4.pdf

Oh, and in general, can anybody tell me why the cathode of the 6x4 is connected to one of the hot leads of the High voltage leads? Never quite understood that..

[richard.cs]

I would be inclined to replace the electrolytics but leave the oil-filled caps. They tend to last forever unless they physically leak oil. Film caps would be a perfectly acceptable replacement I just don't consider them necessary. One change that might improve performance is swapping the OA2's for a zener-string but personally I wouldn''t bother.

If your purpose is to overhaul the supply and make it work as well as it did when do then just go with minimum change, if you want a better performing supply then you'd be best off starting from scratch.

One thing to watch out for if you use film capacitors for high voltages is they tend to exhibit charge recovery much more than electrolytics or oil filled capacitors do. You should make sure that bleeder resistors are present and not open-circuit.

[N2IXK]

For the radiotron designer's handbook, would this be what you're talking about?

http://www.tubebooks.org/Books/RDH4.pdf

Yes, that would be the one referred to. That was "the" classic circuit design manual for the vacuum tube era.

Oh, and in general, can anybody tell me why the cathode of the 6x4 is connected to one of the hot leads of the High voltage leads? Never quite understood that..

That is where the 6X4 gets its AC input voltage. It goes into the cathode rather than the plates, because the 6X4 is being used to develop a negative bias voltage. Both plates are tied together because the tube is being used as a half-wave rectifier.

[calexanian]

Believe me. Put an ac meter in there and you will see much higher ripple than that! The resonance is supposed to be less than 120hz. Far less. It's not a dc filter in a traditional seance. It's to provide an ac component to the current flowing through the choke to provide proper filtering there. You just have to play with those circuits to understand. Look up the section in the RCA radiotron design handbook under power supplies. It's the best description and explanation of how those circuits work anywhere. You can see how they did it before our modern stuff. I highly recommend that book to everybody. Just the basic circuit theory in there is better than any modern textbook.

For the radiotron designer's handbook, would this be what you're talking about?

http://www.tubebooks.org/Books/RDH4.pdf

Oh, and in general, can anybody tell me why the cathode of the 6x4 is connected to one of the hot leads of the High voltage leads? Never quite understood that..

Thats the one. Also. I don't know who started the idea that oil filled caps have high ESR. Tell that to the many thousands of them still being used in discharge welding service! They are still used for big stuff. Radio transmitters, BIG welding equipment. X ray supplies, anything really high voltage. Chicago Condenser is still in business http://www.capacitorindustries.com/ 

Have a look through their website for fun!

[calexanian]

They actually have a foil film replacement for you. have a look at the CMP-805-6C. Looks like an old style one if you want to keep the warm fuzzy feeling of an old oil can.
Might actually have to design something just to get one to play with! BTW. I am not trying to be snarky or obstinate in any way.. Many of these technologies were working well before we were born and will continue working well after we are dust. Just because something is old or deemed to be obsolete, or has a bad name based on feeling rather than science (Leaded glass for example) does not mean it still can have a place or be useful. Don't forget. One of the largest use for tubes still to this day are RF bombarders in the preparation of materials for semiconductors! hahahaha

http://www.capacitorindustries.com/capacitors/high-voltage-dc-capacitors/cmp/default.html